Those days are over when the only use of GPU was to get high frame rate when you play Crysis. For a long time GPUS are used in Graphics Cards to play high-end video games. Initially by developing these cards NVIDIA made their name and their fortune.Many numbers of conceptual technologies depends on the GPU’s ability to parallel run particular algorithms smoothly and fast.
There is a particular area where GPU’s are very useful, and NVIDIA is betting big on it. Today the company is initiating two brand new hardware accelerators, they are also launching a number of tools which will support developers and data managers to handle these accelerators to run deep learning software, they can also use them for photo and video processing.
Today NVIDIA stated that they have joined hands with Mesosphere. Both of their goals is to make it simple for the companies to build and launch accelerated data centers for next generation applications.
Because of Mesosphere’s work with NVIDIA the developers are able to use Apache Mesos, because of it, they will be able to GPU resources in a data center just like we use CPUs and memory. GPU resources will be assembled into a pool and the software autonomously distributes jobs to other machines with compatible GPUs.
As far as the hardware department goes, NVIDIA is going to launch to devices, the M40, and M4 GPU Accelerators. The M40 is designed for machine learning and it is suited for data center use, whereas the M4 is designed for similar work but it consumes low power and it processes video.
Some of well known Public cloud sellers, which include AWS and Microsoft in now GPU centered virtual machines or they are trying to, but the maximum number of data center operators is going for the NVIDIA. Internally Google is betting on machine learning, but they have not been able to those on their cloud platforms yet. But the chances are good that very soon they will be in the field of dedicated machine learning services.
Last Updated on